Fixed-width and decay-based attention mechanisms inspired by working memory improve Transformer grammatical accuracy and human alignment under limited training data.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Working Memory Constraints Scaffold Learning in Transformers under Data Scarcity
Fixed-width and decay-based attention mechanisms inspired by working memory improve Transformer grammatical accuracy and human alignment under limited training data.